Invisible backdoor attack with dynamic triggers against person re-identification (IEEE T-IFS 2023)

This is the pytorch implementation of the paper (accepted by IEEE T-IFS 2023).

Getting Started with DT-IBA

1. Installation:

   • First, install the required packages listed in requirements.txt by running the following command in the command line:

     pip install -r requirements.txt
     

2. Prepare Pretrained Model:

   • The feature extraction phase of our identity hashing network uses a pre-trained SBS-R101. You can obtain it from FastReID's GitHub repository here.
   • The steganography network is referenced from StegaStamp and needs to be retrained according to the image size of the dataset. You can find the StegaStamp GitHub repository here.

   References:

   • FastReID: A Pytorch Toolbox for General Instance Re-identification

     @article{he2020fastreid,
       title={FastReID: A Pytorch Toolbox for General Instance Re-identification},
       author={He, Lingxiao and Liao, Xingyu and Liu, Wu and Liu, Xinchen and Cheng, Peng and Mei, Tao},
       journal={arXiv preprint arXiv:2006.02631},
       year={2020}
     }
     

   • StegaStamp: Invisible Hyperlinks in Physical Photographs

     @inproceedings{2019stegastamp,
         title={StegaStamp: Invisible Hyperlinks in Physical Photographs},
         author={Tancik, Matthew and Mildenhall, Ben and Ng, Ren},
         booktitle={IEEE Conference on Computer Vision and Pattern Recognition (CVPR)},
         year={2020}
     }
     

3. Dataset Preparation:

   • Put the dataset to be poisoned into ./dataset.
   • Set the input and output path, parameters of the pre-trained model, and the configuration of the victim model in backdoor_implantation.py.
   • Run the following command:

     python backdoor_implantation.py
     

Citation:

• If you use the DT-IBA method in your work, please cite this paper:

  @ARTICLE{10285514,
  author={Sun, Wenli and Jiang, Xinyang and Dou, Shuguang and Li, Dongsheng and Miao, Duoqian and Deng, Cheng and Zhao, Cairong},
  journal={IEEE Transactions on Information Forensics and Security}, 
  title={Invisible Backdoor Attack With Dynamic Triggers Against Person Re-Identification}, 
  year={2024},
  volume={19},
  number={},
  pages={307-319},
  doi={10.1109/TIFS.2023.3322659}}